A Quantitative Ultrasound-Based Multi-Parameter Classifier for Breast Masses.

Area of Science:

• Medical Imaging
• Biophysics
• Oncology

Background:

• Distinguishing malignant breast carcinomas from benign fibroadenomas is crucial for patient management.
• Quantitative ultrasound (QUS) offers potential for non-invasive tissue characterization.
• Previous QUS studies often lacked system independence and multi-parameter approaches.

Purpose of the Study:

• To evaluate a model-free, multi-parameter classifier using system-independent QUS parameters for breast tumor differentiation.
• To assess the diagnostic performance of combining two or three QUS parameters in classifying breast carcinomas and fibroadenomas.

Main Methods:

• Radiofrequency ultrasound data were acquired from 43 patients with suspicious breast masses.
• System-independent QUS parameters including specific attenuation (ATT), effective scatterer diameter (ESD), and ESD heterogeneity index (ESDHI) were calculated.
• A minimum Mahalanobis distance classifier was trained and tested using biopsy results, with performance measured by AUC of the ROC curve.

Main Results:

• A two-parameter classifier using ESD and ESDHI achieved an area under the ROC curve (AUC) of 0.98 (95% CI: 0.95-1.00).
• Incorporating specific attenuation (ATT) into a three-parameter classifier (ATT, ESD, ESDHI) further improved performance to an AUC of 0.999 (95% CI: 0.995-1.000).
• The classifier demonstrated high accuracy in differentiating breast carcinomas from fibroadenomas.

Conclusions:

• System-independent QUS parameters, when combined in a model-free classifier, show significant promise for characterizing breast tumors.
• Multi-parameter QUS analysis offers a potentially powerful, non-invasive tool for improving breast cancer diagnosis.
• Further validation in a larger cohort is warranted to confirm these preliminary findings.